News – Thought Leadership
19 February 2021
ARIA – the Advanced Research and Invention Agency – is the new £800 million scientific research agency announced by Business Secretary Kwasi Kwarteng. It draws its inspiration from the US Defense Advanced Research Projects Agency (DARPA). It is designed to be unbureaucratic, to “embrace failure” and to invest in “high-risk, high-reward” projects.
The principle of ARIA is very welcome. Government funding is often inherently risk averse; civil servants are not traditionally instructed to ‘embrace failure’. A ‘small-c’ conservative approach to funding R&D, misses the most radical innovation and ideas. Too much funding is channelled towards subsidising the incumbents rather than finding the plucky upstarts. This is not good for diversifying innovation and it is not good for finding the next breakthrough technology.
The US DARPA embraces failure by showering huge amounts of funding on a swathe of innovators and ideas and then brutally cutting funding when things fail – by taking this approach they occasionally turn up pearls like the internet or driverless cars. DARPA is directly linked to the US Defense Department’s long-term strategy and procurement plans, which gives it a clear focus.
ARIA on the other hand, will have a much more limited budget and is not directly linked to the procurement goals of a government department. That means there is potentially less focus and also less room for finding the occasional breakthrough in a sea of celebrated failure.
One of the main findings from the Science and Technology report into the new agency rightly recommended focusing on central missions and linking these missions with government departments.
ARIA can achieve its laudable goals, but only if it is sufficiently focused on areas where its more possible to predict likely pearls, and only if it is then selective about the innovation funding methods it uses to find them.
You can legitimately take bigger risks with ARIA only if you take greater care. The DARPA model has highly autonomous programme leads who are empowered to make big decisions about funding, without extensive oversight, because their approach is expert-led, data-led and they do their due diligence.
ARIA will require a strong capability in predicting potential technology pathways that combines a strong analytical approach to identifying where good innovation is coming from with a capability to think creatively about the future.
Kwasi Kwarteng has said that ARIA will experiment with funding models including prize incentives. This is also welcome.
The efficacy of traditional grants are limited in areas where there is high uncertainty. That’s because they fund activities rather than outcomes. This discourages the charge to ‘embrace failure’ because funding is given upfront. It is essentially an educated gamble and that means that stakes are more likely to be placed on those with more predictable trajectories.
One approach used by DARPA that should be deployed here is to combine grants with outcome based funding such as challenge prizes.
Challenge Works has published the Great Innovation Challenge which provides examples of challenge prizes used in the US including by DARPA and suggests areas of focus for breakthrough challenge prizes in the UK.
Another approach is to intelligently cultivate innovation with non-financial incentives – preferential access to regulators or high value data sets – as good examples.
The intention is that ARIA will be agile and free from bureaucratic impediments. This is the right instinct.
The recent NAO report of the Industrial Strategy Challenge Fund found that the government took 72 weeks to approve the challenges, and then a further 31 weeks to assess applications. Two years is a lifetime for some fast moving new technologies and ARIA needs to be much more agile. This is the driver to exempt ARIA from Freedom of Information laws, but this will backfire.
Radical openness and honesty is needed or distrust will undermine it. The public will expect to know what’s happening with public money and greater risk requires transparency and evaluation in order to determine what works.
ARIA is a welcome bold step, the instincts are good, but it will thrive or fail according to the methods it uses.
